Biomedical Engineering Reference
In-Depth Information
Table 13.1:
Median error for CT-to-MR registration in 'mm' between the
prospective gold-standard and several retrospective registration techniques.
The label “rect.” indicates that the MR image was corrected for geometrical
distortion before registration. The result of the technique using the
binarization approach is labelled LO1, while the one using the nonlinear
binning method is labelled LO2.
BA
CO
EL
HA
HE
HI
MAI
MAL
NO
PE
RO1
LO1
LO2
CT-T1
1.6
1.5
1.6
3.4
1.4
1.2
5.1
4.3
3.3
2.7
4.2
1.26
1.24
CT-PD
1.9
1.5
2.0
3.1
2.4
1.9
4.1
4.0
7.8
1.9
4.5
1.67
1.90
CT-T2
2.5
1.5
1.6
4.2
4.7
1.5
3.9
5.0
3.9
2.5
4.5
1.64
1.47
CT-T1-rect.
1.4
0.7
0.9
3.3
1.0
0.7
4.9
5.4
3.4
2.2
5.9
0.65
0.96
CT-PD-rect.
1.7
0.8
1.1
3.0
1.7
0.7
3.0
4.0
4.6
2.1
5.9
0.85
0.90
CT-T2-rect.
2.1
0.8
1.6
3.5
1.6
0.8
4.3
5.3
4.2
2.9
5.5
0.81
0.89
Table 13.2:
Maximum error for CT-to-MR registration. See notes in Table 13.1.
BA
CO
EL
HA
HE
HI
MAl
MAL
NO
PE
RO1
LO1
LO2
CT-T1
6.4
6.7
6.0
51.8
11.0
2.8
12.8
61.4
10.4
7.3
26.0
3.15
2.76
CT-PD
6.9
3.6
6.6
49.6
10.4
4.1
19.0
59.0
13.9
4.3
25.9
3.64
3.91
CT- T2
9.1
3.4
4.1
50.6
13.6
4.2
6.3
59.5
9.7
7.2
26.7
3.64
4.64
CT-T1-rect.
5.8
3.8
2.6
48.2
2.1
2.3
14.2
60.9
9.6
5.9
27.8
1.98
1.95
CT-PD-rect.
5.9
2.5
5.3
45.9
3.7
2.3
9.9
62.7
11.5
4.6
27.5
2.13
1.81
CT-T2-rect.
7.4
4.3
5.2
49.1
14.3
3.0
6.5
63.2
10.2
9.0
27.1
2.65
2.05
Table 13.3:
Number of optimization steps used by each of the three
binning techniques for first level.
Binarization
Linear binning
Nonlinear binning
Data Set
Mean
δ
Mean
δ
Mean
δ
Patient 1
98
21.85
165
51.66
139
17.72
Patient 2
140
22.98
244
65.85
208
56.72
Patient 3
128
17.51
244
66.02
251
23.25
Patient 4
163
38.94
246
68.41
211
50.42
Patient 5
151
42.66
195
24.28
236
52.2
Patient 6
153
32.69
199
53.76
268
56.55
Patient 7
112
27.12
228
67.98
226
71.77
